Upload
trinhduong
View
222
Download
1
Embed Size (px)
Citation preview
Optimal Fiscal Policy with Consumption Taxation∗
Giorgio Motta† Raffaele Rossi‡
August 2016
Abstract
We characterise optimal fiscal policies in a tractable Dynamic General Equilibrium model with mo-
nopolistic competition and endogenous public spending. The government has access to consumption
taxation, in addition or as alternative to labour income taxation. First we find that the optimal share
of government spending is higher with consumption taxes than with labour income taxation. Then
we show that using consumption taxation is welfare superior as this tax component acts as indirect
taxation of profits (intratemporal gains of taxing consumption) and enables the policy-maker to manage
the burden of public debt more efficiently (intertemporal gains of taxing consumption). We quantify
each of these welfare gains by calibrating the model on the US economy.
JEL classification: E62, H21.Keywords: fiscal policy, consumption taxation, endogenous government spending.
1 Introduction
What are the benefits from taxing consumption? We answer this question by studying Ramsey fiscal
policies in a tractable and deterministic dynamic general equilibrium model that abstracts from capital
accumulation, where firms have monopolistic power, public spending directly increases households’ utility
and the government balances its budget by levying distortionary linear labour income and consumption
taxes (but no lump-sum taxes) and by issuing risk-less bonds.
∗We are grateful to Emanuele Bracco, Stefano Gnocchi, Sarolta Laczo, Dmitry Matveev, Monika Mertz, and MaurizioZanardi, Michela Cella, Andrea Colciago, Davide Debortoli and Nick Snowden, as well as seminar participants at the Bankof England, University of Milano-Bicocca, University of Durham, University of Vienna, University of Liverpool, Universityof Lancaster, University of Glasgow, the 2014 T2M conference in Lausanne, the 2013 CEF conference in Vancouver, the 2014EEA meeting in Toulouse and the 2015 RES conference in Manchester.†Department of Economics, Lancaster University Management School, LA1 4XY, Lancaster, United Kingdom. Email:
[email protected].‡Department of Economics, University of Manchester, Manchester, United Kingdom. Email: raf-
1
Compared to a scenario where only labour income taxes are available, we identify and quantify two
benefits of taxing consumption. First, consumption taxation serves as indirect taxation of inefficient prof-
its. We call this the intratemporal gains of taxing consumption. These intratemporal gains increase with
the monopolistic power of firms and disappear under perfect competition. Second, consumption taxation
helps to manage more efficiently the price of public debt, via direct manipulations of the households’
Euler equation. We call this the intertemporal gains of taxing consumption. These intertemporal gains
grow with the initial level of public debt and vanish when the initial level of debt is zero. In a numerical
exercise aimed to replicate some salient features of the US economy, we find that both gains are quan-
titatively important. These gains hold under some important extensions of the baseline model, such as
non-separability between leisure and consumption and complementarity-substitutability between private
and public spending.
Furthermore we show that compared to a scenario where only labour income taxes are available, these
two characteristics of consumption taxation allow the policy-maker to set a higher government spending-
to-income ratio. For the special case of separable preferences, we are able to present a set of analytical
conditions on the optimal provisions of public goods. In particular, if households hold logarithmic pref-
erences in private and public consumption, the optimal share of public-to-private consumption is at the
efficient level under consumption taxes, while the same ratio is lower under labour income taxation, and
it gets smaller the higher the monopolistic power of firms.
From this, the main contributions and novelties of this paper lay in quantifying and disentangling the
different gains from taxing consumption and from presenting a set of analytical conditions on the optimal
provision of public spending. Our model sacrifices complexity so that our results may be conveyed
transparently, and so that we may obtain analytical results.
This paper links to the existing literature in several ways. First, we contribute to the literature that
studies the welfare gains of taxing consumption. The closest contribution can be found in Coleman (2000),
2
who finds, in a deterministic model with capital accumulation and perfect competition, that replacing
income taxes with consumption taxes would lead to large welfare gains in the United States. Correia
(2010) extends this result to a heterogeneous-agents framework. Two recent contributions highlight the
role of consumption taxation as a tool to relax a constraint of the monetary authority on the nominal
interest rate, either as a result of the zero lower bound (Correia, Farhi, Nicolini, and Teles, 2013) or in a
monetary union (Farhi, Gopinath, and Itskhoki, 2014). However none of these contributions highlights and
quantifies the interactions of consumption taxation with monopolistic power, public debt and government
spending.
Second, we contribute to the literature on the optimal provision of public goods. Adam (2011) analyses
optimal government spending in a New Keynesian model with labour income taxation and finds that, as
here, public spending-to-income ratio fall short of its first best level. Klein, Krusell, and Rıos-Rull (2008)
study the provision of public spending in a Real Business Cycle model with labour income taxation when
the policy-maker may or may not be able to commit to future policies. They find that the optimal share
of public spending-to-income is higher under commitment. However this literature does not consider the
role of consumption taxation in shaping the optimal provision of public good.
The paper is organized as follows. Section 2 sets up the model. Section 3 presents the policy analysis.
Finally, Section 4 concludes.
2 Model
We present a flexible price economy with linear technology in labour, without uncertainty nor capital
accumulation, where firms have monopolistic power. The benevolent government finances an endogenously
determined level of government spending (gt) through a proportional labour income tax (τht ∈ (−∞, 1)),
a linear consumption tax (τ ct ∈ (−1,∞)) and by issuing a one-period discount bond (bt+1). Its budget
3
constraint is
gt + bt = τ ct ct + τht wtht + pbtbt+1, (1)
where ct is private consumption, wt are real wages, ht represents labour supply and pbt is the ex-ante real
bond price.
The representative household takes prices and policies as given, chooses consumption, savings via
public bond holding (bt+1), and leisure (`t), with the standard time constraint, 1 = ht + `t, in order to
maximise the expected discounted value of her lifetime utility, i.e.
maxct,`t,bt+1
∞∑t=0
βtu (ct, `t, gt), (2)
where β ∈ (0, 1) is the discount factor. The utility function in (2) is separable across time, twice con-
tinuously differentiable, with a constant Frisch elasticity of labour supply. Furthermore we assume that
utility is increasing in its three arguments, that is uc,t > 0, u`,t > 0 and ug,t > 0, concave, i.e. ucc,t < 0,
u``,t < 0 and ugg,t < 0, and that allows for non-separability between leisure and consumption and comple-
mentarity/substitutability between private and public spending. We indicate respectively with ux,t and
uxx,t the first and the second partial derivatives of ut with respect to the generic variable x at time t.
Both private and public consumption baskets consist of constant-elasticity of substitution aggregators
of individual goods i, i.e. ct =
(∫ 10 c
ε−1ε
it di
) εε−1
and gt =
(∫ 10 g
ε−1ε
it di
) εε−1
, where ε > 1 is the elasticity of
substitution. The household’s budget constraint can be written as
(1 + τ ct ) ct + pbtbt+1 = wt (1− `t)(
1− τht)
+ bt + dt, ∀t, (3)
where dt are the aggregate profits rebated from the monopolistic competitive producers. The household’s
optimal behaviour is characterised by the following first order conditions
u`,tuc,t
= wt
(1− τht
)(1 + τ ct )
, ∀t, (4)
and
pbt = βuc,t+1 (1 + τ ct )
uc,t(1 + τ ct+1
) , ∀t, (5)
4
Examining the household’s first-order conditions, the different distortions caused by the two tax
instruments become apparent. The labour income tax distorts the (intratemporal) consumption-leisure
margin, (4). The current consumption tax distorts the same margin. In addition, both the current and
next period’s consumption tax enters into the current (forward-looking) Euler equation, (5).
We now present the supply side of the economy. Each of the differentiated goods i ∈ [0, 1] is produced
by one monopolistic firm who employs labour using a linear production technology yi,t = hi,t, where yi,t
is the output of the generic firm i. Via standard optimality conditions, one can show that the marginal
costs for firm i are simply equal to the real wages wt. Monopolistic competition implies that firms can
charge a price higher than their marginal costs. Normalising prices to 1, this implies µiwt = 1, where
µi = εε−1 ≥ 1 represents the (time-invariant) mark-up of firms i over their marginal costs. Finally, firm’s
i profits are equal to di,t = (1− wt) yi,t = 1εyi,t. Note that under perfect competition (ε→∞) profits will
be zero. We concentrate on symmetric equilibrium. Thus the following aggregations hold for any i, 1)
hi,t = ht; 2) yi,t = yt; 3) µi,t = µt; and 4) di,t = dt. We are now ready to provide a formal definition of
the decentralised equilibrium.
Definition 1 (Decentralised Equilibrium). A decentralised equilibrium consists of government policies,{τht , τ
ct , gt, bt+1
}∞t=1
, prices,{wt, p
bt
}∞t=1
, and private sector allocations, {ct, `t, ht, yt, dt}∞t=1, satisfying,
the government’s budget constraint, (1); the private sector optimisation taking government policies and
prices as given, that is, the household’s time constraint, `t + ht = 1, budget constraint, (3), and op-
timality conditions, (4), (5); the market clearing conditions, ct + gt = yt; the transversality condition
limT→∞ βT uc,T1+τcT
bT = 0 and the no-Ponzi game condition limT→∞
[(∏T−1i=0 p
bi
)bT
]= 0, taking the initial
level of debt b0 as given.
3 Ramsey Policy
Our policy exercise starts by defining the efficient allocation of the model.
5
Definition 2 (First Best). The Social Planner’s Program defines the first-best allocation. This consists
of allocations {gt, ct, `t, ht, yt}∞t=0 that maximise (2) subject to household’s time constraint, 1 = ht + `t,
the production function, yt = ht, and the market clearing condition, ct + gt = yt.
Proposition 1 (Social Planner’s Problem). The Social Planner’s allocation is
uspc,t = usp`,t = uspg,t, ∀t. (6)
Proof. The proof is straightforward. First let us substitute the production function into the market
clearing condition to eliminate yt and have a single constraint to the maximisation problem. Then by
taking the first order conditions and combining them through the Lagrangian multiplier, we obtain (6).
Where a variable with the superfix sp indicates its level in the Social Planner (or first best) allocation.
In the first best equilibrium the marginal utilities of private and public consumption must equate the
marginal utility of leisure. This simple allocation rule is optimal because it is equally costly to produce
public and private consumption goods.
Definition 3 (Ramsey Problem). The Ramsey policy-maker maximises (2) over the decentralised equi-
libria. A Ramsey outcome is a decentralised allocation that attains the maximum of (2).
The first natural question arising in our policy exercise is whether or not the set of fiscal instruments{τht , τ
ct , gt, bt+1
}∞t=1
suffice the Ramsey policy-maker to decentralise the first-best allocation in (6), for
any exogenous level of b0. The answer is no. As shown more formally in Motta and Rossi (2013), the
decentralisation of the efficient allocation requires i) at least one tax instrument between τht and τ ct to be
a subsidy, i.e. to be negative; and ii) large government asset positions, i.e. negative b0. Put it differently,
the Ramsey Planner cannot obtain the first best allocation with a generic initial level of public debt b0, as
this would imply either τ ct < −1 or τht > 1. Condition i) is necessary to guarantee that the marginal rate
of substitution between leisure and consumption equals 1, while condition ii) is necessary to satisfy the
6
government budget constraint. The reason for this result is the following. Consider, for sake of simplicity,
a simplified model with no public spending, no public debt and perfect competition (ε → ∞, so that
wt = 1 ∀t). This model is static, the market clearing condition is ht = ct and consumption and labour
income taxes have the same equilibrium tax base. Then, the decentralization of the first best implies
τ ct = −τht ∀t. This tax policy would raise zero revenues in equilibrium, i.e. |τ ct ct| = | − τht ht|. However,
in the presence of monopolistic competition, i.e. ε <∞, full efficiency implies −τht = 1ε−1 + τ ct
εε−1 as the
policy-maker needs to compensate households for the fact that the real wage falls short of equating the
marginal rate of substitution between consumption and leisure. Given that(
1ε−1 + τ ct
εε−1
)h > τ ct ct, this
policy implies negative revenues in equilibrium.
Moreover, using numerical solutions which we describe in detail below, we find that for any given level
of b0, the policy-maker always finds a marginal welfare advantage in increasing consumption taxation
and subsiding labour. This means that a potentially revenue-neutral combination of an increase in the
consumption tax and a decrease in the labour income tax, always increases efficiency. However, while
this policy will never replicate the first best allocation, it would create extreme tax and subsidy positions.
Hence, this policy would be extremely difficult to implement due, for example, to the high costs associated
in verifying hours worked. Similarly, a very high consumption tax rate would perhaps lead to a significant
amount of unreported barter. Therefore, we study and compare the cases where the Ramsey Planner is
constrained in using only one tax instrument, i.e. either τ ct = 0 or τht = 0. This is common practice in the
literature when optimal policy implies extreme tax positions, e.g. Coleman (2000), Correia (2010) and
Martin (2010). In what follows we are going to express the Ramsey problem with a primal approach, i.e.
we write the policy problem solely in terms of allocations rather than tax rates.
Proposition 2 (Ramsey Problem with Labour Income Taxes). For any initial level of debt b0, the Ramsey
policy-maker chooses allocations {ct, `t, gt}∞t=1 and policies {bt+1}∞t=1 in order to maximise (2) subject to
7
the feasability constraint 1− `t = ct + gt and the implementability constraint
uc,tct + βuc,t+1bt+1 = u`,t (1− `t) + uc,tbt + uc,t1
ε(1− `t) , ∀t = 0, 1, 2, .... (7)
Proof. The implementability constraint (7) is obtained by substituting into (3) taxes and real wages via
(4) and bond prices via (5) and then imposing τ ct = 0.
Taking derivatives to the associated Lagrangian problem, the optimality conditions are
u`,t (1 + γt)− γt[u``,t (1− `t)−
1
εuc,t
]+ uc`,t
{γt
[ct −
1
ε(1− `t)
]− (γt − γt−1) bt
}= λt, (8)
uc,t (1 + γt) + γtucc,t
[ct −
1
ε(1− `t)
]− (γt − γt−1)ucc,tbt − γtu`c,t (1− `t) = λt, (9)
ug,t + γt
[ucg,tct − u`g,t (1− `t)− ucg,tbt − ucg,t (1− `t)
1
ε
]+ γt−1ucg,tbt = λt, (10)
γt+1 = γt, γ−1 = 0, (11)
where uxy,t indicate the cross derivatives of the generic variables x and y at time t and λt ≥ 0 and
γt ≥ 0 denote the Lagrange multipliers attached to the feasibility and the implementability constraints,
respectively. Importantly, (11) implies that γt must be constant ∀t > 0. As presented in Lucas and
Stokey (1983), in this class of models the steady-state level of debt is not determined. Given that in the
equilibrium γt must be constant, all the remaining conditions are identical ∀t ≥ 1, and so do the values
of c, g, `, and b. Thus, the steady-state allocations depend on γ. Equation (9) evaluated at time 0 shows
that γ depends on the initial level of debt, b0. Hence the steady-state of the Ramsey Problem depends
on initial conditions.
Next we set out the Ramsey problem for the case of consumption taxation.
Proposition 3 (Ramsey Problem with Consumption Taxes). For any initial level of debt b0, the Ramsey
policy-maker chooses allocations {ct, `t, gt}∞t=1 and policies {bt+1}∞t=1 in order to maximise (2) subject to
the feasability constraint 1− `t = ct + gt and the implementability constraint
uc,tε− 1
εct + βu`,t+1bt+1 =
ε− 1
εu`,t (1− `t) + u`,tbt + u`,t
1
ε(1− `t) (12)
8
Proof. See Proof to Proposition 2.
In this case, the first order conditions of the Ramsey problem with consumption taxation can be
written as
ug,t + γtucg,tε− 1
εct − (γt − γt−1)u`g,tbt − γtu`g,t (1− `t) = λt (13)
u`,t + γtucg,tε− 1
εct − (γt − γt−1)u``,tbt − γt (1− `t)u``,t + γtu`,t = λt (14)
uc,t + γtε− 1
ε(ucc,tct + uc,t)− (γt − γt−1)u`c,tbt − γtu`c,t (1− `t) = λt (15)
γt+1 − γt = 0 γ−1 = 0 (16)
As in the case of labour income taxes, the steady-state of the Ramsey Problem depends crucially on the
initial level of debt b0. Similarly, the transition from the initial conditions to this steady-state lasts one
period. However, under the two different tax instruments, the Ramsey Planner faces different imple-
mentability constraints, i.e. (7) versus (12). This implies that, given the same initial level of public debt
b0, and the structural parameters of the economy, the allocations along the transition and in turn the
welfare properties of the two policy scenarios will be in general different. The aim of this policy exercise
is to explore these differences.
3.1 Analytical Results
We first impose some special conditions to the model that allow us to present a set of analytical results.
These results will be instructive to develop transparent economic intuitions. Let us assume that the
Utility Function in (2) is separable in its three arguments, i.e. ucg,t = u`g,t = u`c,t = 0 and that displays
log preferences in private and public consumption, i.e. −ucc,tct = uc,t and −ugg,tgt = ug,t.
Proposition 4 (Optimal Government Spending with Labour Income Taxation). Under labour income
taxation, given a generic level of initial outstanding debt b0, the Ramsey Planner sets
ug,t ≥ u`,t and ug,t ≥ uc,t, ∀t ≥ 1, (17)
9
Where the function Gt ≡ ug,t−uc,t, is monotonically decreasing in ε, the parameter governing monopolistic
power. Under perfect competition, ε → ∞, Gt = 0, so that the share of government spending to total
income is at the efficient level.
Proof. See Appendix.
This proposition shows that it is optimal in the presence of monolistic competition to set public
consumption below its efficient level . This is because there exists a wedge between the marginal utility
of leisure and the marginal utility of consumption, which is composed of two components. First, the
monopolistic power that firms hold. Second, the distortive nature of fiscal policy used to finance public
spending and public debt. By reducing government spending, the Ramsey Planner can lower the tax
rate thus shrinking the wedge between consumption and leisure. Consider, for instance, any increase in
inefficiency, due, for example, to more market power. This would dampen economic activity. The Ramsey
Planner by adjusting the government size relative to GDP and therefore the level of taxation, can affect
households labour supply and private consumption and potentially reduce inefficiency.
Interestingly, under perfect competition, i.e. ε → ∞, the optimal steady-state ratio of public to
private consumption would be equal to the one chosen in the efficient allocation, even though labour
supply is distorted by the income taxes and indirectly by public debt. It should be noted that this result
does not hold in general and depends, in particular, on the utility function that implies under perfect
competition that income and substitution effects of taxation on labour supply cancel each other out. In
turn, when ε < ∞, the extra inefficiency represented by untaxed profits makes the substitution effect
of labour supply stronger, thus inducing the Ramsey Planner, other things equal, to reduce the public
spending-to-income ratio. For this reason, the logarithmic utility function in private consumption we have
chosen is particularly instructive, as it allows to isolate in a transparent way how the interaction between
of monopolistic competition and labour income tax affects the optimal provision of public consumption.
We now turn our attention to the case when the Ramsey Planner can access consumption taxation.
10
Proposition 5 (Optimal Government Spending with Consumption Taxes). Under consumption taxation,
given a generic level of initial outstanding debt b0, the Ramsey Planner sets
ug,t = uc,t, ∀t ≥ 1, (18)
and the share of government spending to total income is as in the Social Planner equilibrium.
Proof. See Appendix.
Under consumption taxation, the Ramsey planner chooses to fix the share of government spending
to private consumption as in the efficient allocation independently of the level of long-run public debt
or monopolistic power in the economy. This result is driven by the specific shape of the utility function
and by the nature of consumption taxation. The log specification of the utility function guarantees
that for any change in consumption tax rate, triggered for instance by higher long-run public debt, the
substitution and income effect of labour supply are of the same magnitude. At the same time, any change
in monopolistic power is absorbed by consumption taxation, which acts as indirect taxation of profits.
This latter mechanism is crucial in order to understand the optimal provision provision of public spending
and the efficiency gains from taxing consumption. Furthermore, by inspecting Propositions 4 and 5, it
results apparent that for any level of long-run level of public debt and monopolistic power, the optimal
share of government spending to total output is higher under consumption taxation than under labour
income tax.
3.2 Quantitative Analysis
We now turn to numerical methods to find the optimal fiscal policy mix quantitatively with and without
consumption taxation. We will also quantify the welfare gains from taxing consumption. In order to see
how the allocations and tax rates behave, we solve a calibrated version of our economy. In what follows
we specify the benchmark calibration of the model as well as some alternative parametrisations. To this
end, we need to make some further assumptions up front. We specify the utility function as,
11
u (ct, `t, gt) =(ct + αgt)
1−σ(
1− κ (1− σ) (1− `t)1+1/ϕ)σ− 1
1− σ+ ωg
g1−σgt
1− σg, for σ 6= 1, σg 6= 1, (19)
and
u (ct, `t, gt) = log (ct + αgt)− κ (1− `t)1+1/ϕ + ωglog (gt) , for σ = 1, σg = 1. (20)
where ϕ is the (constant) Frisch elasticity of labour supply and κ is the utility weight of leisure. The
parameter α indicates the degree to which government and private consumption are substitute in the
Edgeworth sense. That is, the marginal utility of private consumption increases (decreases) with public
spending if α is negative (positive). The term ωgg1−σgt1−σg allows complementarity (i.e., α < 0) and avoids
that the representative household receives a negative utility stream from government consumption, where
ωg is the utility weight on public goods and σg is the curvature of the utility in g. The parameter σ ≥ 0
is the inverse of the constant elasticity of intertemporal substitution (EIS) in the case of α = 0, while for
the general case of α 6= 0 the elasticity of intertemporal substitution is also function of private and public
consumption. On the same note, when α = 0, our utility function features a constant income effect of
labour supply that is independent from the EIS and it is equal to 1. The model is calibrated at annual
frequency on US data over the period 1996-2010. Data are collected from Trabandt and Uhlig (2011).
We set the discount factor β to 0.979. This value implies an average annual bond prices of 2.2%. For
the benchmark case, we set σ and σg equal to 1, so that our benchmark utility specification specialises
to (20). We then allow σ parameters to vary in the range [0.5, 2], which covers most macro and micro
literature on the estimation of the EIS. Given an intertemporal elasticity of substitution equal to 1, we set
the Frisch elasticity of labour supply equal to 1 in our benchmark case, a value consistent with Kimball
and Shapiro (2008). The micro and macro literature tend to differ on the estimates of the Frisch elasticity.
For this reason we also control the robustness of our results with a wide range of values of ϕ. We calibrate
κ such that, given the other parameters, the optimal supply of labour in the efficient equilibrium is the
30% of the time endowment, i.e. ` = 0.7. We choose ωg such that in the efficient equilibrium the ratio
12
of government spending to GDP in t = 0 is 20%. In the benchmark case we set α = 0, then we allow to
vary it in the range of (−0.5, 0.5), consistently with the estimations of Ni (1995) and the parametrisation
of Natvik (2009). We calibrate the elasticity of substitution among good varieties ε to 6 which implies a
price mark-up of around 20%, a value consistent with the data, see for example Rotemberg and Woodford
(1998). Given how crucial this parameter is in shaping the optimal government spending under the two
tax scenarios, we allow this elasticity to vary between 5 and ∞, that correspond to the high mark-up
(ε = 5) and to perfect competition (ε =∞) cases. Finally, in our benchmark case we fix the initial level
of public debt, b0 so that the share of debt-to-GDP, γb in t = 0 equals 0.64. We then solve for the optimal
allocation for different values of debt-to-GDP ratios. The complete description of the parametrisation
adopted is reported in Table (1).
As explained in Section 3, the Ramsey problem can be recast as a two-period problem, at t = 0 and at
t ≥ 1, imposing the extra condition that γ−1 = 0. By doing so, one can find the solution to the Ramsey
allocation solving a system of non-linear equations and time 0 and for any t ≥ 1, given the initial level of
debt, b0.1
Table 1: Calibrated Parameters
Parameter Benchmark Alternative Description Restriction
ϕ 1 0.2-5 Frisch Elasticity Literatureσ 1 0.8-2 IES Literatureσg 1 - Curvature Public Spending Literatureβ 0.979 - Discount Factor Dataκ 6.944 Various Weight of Leisure Dataωg 0.25 Various Weight of Public Goods Dataα 0 -0.5-0.5 Edgworth Parameter Literatureε 6 5-∞ Elasticity of Substitution Dataγb 0.64 0-1.2 Debt-to-GDP Ratio Data
Table (2) reports the allocations and welfare under the benchmark calibration for the cases of labour
income tax and consumption tax, as well as the efficient allocation. A standard result in the optimal
taxation literature shows that a Ramsey government has the temptation to influence favourably the bond
prices pbt given by the Euler equation (5), see Lucas and Stokey (1983) among others. This is the time
1We try several solution algorithms (non-linear solvers) and all deliver the same results.
13
inconsistency problem that affects the Ramsey policy. In a generic period t ≥ 1 current consumption
influences both pbt and pbt−1. As a consequence, if the government uses taxes and public expenditure to
increase the price of the bond pbt , other things equal, pbt−1 decreases. Instead at t = 0 consumers’ savings
and previous prices (pb−1) are given. Therefore, if the government inherits a positive level of debt, it can
benefit from an increase in the price of the bond (pb0) without incurring any additional cost.
Under labour income taxes, this incentive to manipulate bonds prices translates into an increase in
initial private consumption c0, thus fostering the demand for saving, and allowing the government to sell
its bonds at a more convenient price. This policy is implemented by reducing τh0 below its long run level
thus increasing debt and debt-to-income ratio at t ≥ 1. All in all, the resulting lower interest rate allows
the government to sell its bond at a higher price. At the same time, lower taxes in the initial period
push households to postpone leisure. As a result, income in the initial period is above its long run level.
Finally and consistently with Proposition 4, the public spending income ratio is below the efficient level
along the entire transition path.
Under consumption taxation the incentive to manipulate the initial bond prices is lower, as the Ramsey
planner can directly boost the initial level of savings via an increase of the tax rate in the initial period. In
turn the time inconsistency problem under consumption taxation is less severe than under labour income
taxation. This policy allows to reduce debt and debt-to-income ratio for any t ≥ 1, thus reducing the
need of fiscal revenues in the long run. Higher taxes in the initial period create an incentive to postpone
labour supply and consumption. As a result, income is lower in t = 0 than in the subsequent periods.
As shown analytically in Proposition 5, under consumption taxation the government spending to income
ratio is at the efficient level for the entire transition path.
The dramatic policy differences between the two tax schemes (labour income vs consumption) imply
a substantial welfare gain from taxing consumption. It is interesting to note that the gains from taxing
consumption come both from the role that this taxation has in indirectly taxing profits, thus reducing
14
inefficiency, (intratemporal gains from taxing consumption) and from the dynamic properties of consump-
tion taxation which allows to reduce debt in the long run (intertemporal gains from taxing consumption).
In our benchmark calibration, despite the fact that the consumption tax base is smaller than the labour
income tax base, taxing consumption rather than labour income can reduce our measure of welfare loss
by 22.3%.2
Table 2: Tax Rates and Allocations, Benchmark Case
First Best τ c = 0 τh = 0Variable t = 0 t ≥ 1 t = 0 t ≥ 1
Consumption tax rate - 0 0 0.400 0.263Labour income tax rate - 0.074 0.247 0 0
Hours worked, Income 0.300 0.257 0.236 0.231 0.244Leisure 0.700 0.743 0.764 0.768 0.756Consumption 0.240 0.216 0.191 0.185 0.195Public spending 0.060 0.042 0.045 0.046 0.049Public debt - 0.165 0.172 0.148 0.117Public Debt Price - 1.104 0.979 1.031 0.979
Consumption-income ratio 0.800 0.838 0.810 0.800 0.800Public spending-income ratio 0.200 0.162 0.190 0.200 0.200Public debt-income ratio - 0.640 0.730 0.640 0.479
Welfare-eq. consumption loss 0 0.063 0.049
In Table (3) we solve the model with different degrees of monopolistic competition.3 In order to do
so we vary the parameter ε. For the high monopolistic case we set ε = 5, a value that implies a mark-up
of 25%. Then we solve the model under perfect competition, i.e. ε → ∞.4 This exercise is particularly
instructive as it allows to gather a better insight about the welfare gains of taxing consumption, i.e.
intratemporal vs. intertemporal gains. This is because the role of consumption taxes as indirect taxation
of profits disappears under perfect competition. Hence the resulting welfare gains of taxing consumption
2Welfare is calculated in the following way. Let {cspt , `spt , g
spt }∞t=0 be the allocation in the efficient equilibrium and{
cit, `it, g
it
}∞t=0
the allocation under the generic policy i. The last row in Table 2 reports the percentage increase in con-sumption υ making the utility in the first best allocation equivalent to the alternative utility under policy i, i.e.
∞∑t=0
βtu(cit ∗ (1 + υi), `it, g
it
)=
∞∑t=0
βtu (cspt , `spt , g
spt ) ,
while the welfare gains of taxing consumption is given by(
1− υct
υit
), where υit and υct are the consumption equivalent
welfare losses under labour income and consumption taxation, respectively.3Along the robustness analysis, for brevity reasons, we use the log specification of the utility function. In an Online
Appendix available at this link, we solve the model with several other combinations of parameters.4In order to mimic the perfectly competitive case, we set ε = 104.
15
are generated only by the intertemporal dimension.5 In our parametrisation, the intertemporal gains of
taxing consumption amount to 3.70%. Differently when monopolistic power is high, i.e. ε = 5, the role of
consumption taxation as an indirect profits tax magnifies (intratemporal gains from taxing consumption).
Hence the welfare gains from taxing consumption increases with respect to the benchmark calibration and
amount to 24.38%. Moreover it is interesting to see that under labour income tax the share of government
spending to total output decreases respect to the benchmark case. As explained in Proposition 4, this
is due to the negative substitution effect on labour supply generated by the increase in monopolistic
competition which pushes the Ramsey planner to reduce public spending.
Table 3: Tax Rates and Allocations, Monopolistic Power
First Best τ c = 0 τh = 0Variable t = 0 t ≥ 1 t = 0 t ≥ 1
ε→∞, κ = 6.944, ωg = 0.25Consumption tax rate - 0 0 0.349 0.264Labour income tax rate - 0.130 0.215 0 0Hours worked, Income 0.300 0.277 0.266 0.258 0.267Leisure 0.700 0.723 0.734 0.742 0.733Consumption 0.240 0.226 0.213 0.206 0.213Government Spending 0.060 0.051 0.053 0.052 0.053Public debt - 0.177 0.185 0.165 0.143Public debt price - 1.041 0.979 1.011 0.979Consumption-income ratio 0.800 0.816 0.800 0.800 0.800Public spending-income ratio 0.200 0.184 0.200 0.200 0.200Public debt-income ratio - 0.640 0.695 0.640 0.537Welfare-eq. consumption loss 0 0.017 0.016
ε = 5, κ = 6.944, ωg = 0.25Consumption tax rate - 0 0 0.410 0.262Labour income tax rate - 0.060 0.253 0 0Hours worked, Income 0.300 0.253 0.230 0.226 0.239Leisure 0.700 0.747 0.770 0.774 0.761Consumption 0.240 0.214 0.187 0.181 0.191Government Spending 0.060 0.040 0.043 0.045 0.048Public debt - 0.162 0.169 0.145 0.112Public debt price - 1.119 0.979 1.035 0.979Consumption-income ratio 0.800 0.844 0.813 0.800 0.800Public spending-income ratio 0.200 0.156 0.187 0.200 0.200Public debt-income ratio - 0.640 0.736 0.640 0.468Welfare-eq. consumption loss 0 0.077 0.058
Table (4) presents the results from solving the model for different initial level of debt. This exercise
is important, because by varying b0 and keeping the monopolistic competition at its benchmark value,
5With perfect competition ε→∞, and no public debt γb = 0, the welfare gain from taxing consumption is zero.
16
we can control and in turn shut down the intertemporal gain of taxing consumption. In the limiting
case with no initial debt, the Ramsey problem reduces to a static maximisation problem. This case is
presented in the top quadrant of Table (4). Here the gains from taxing consumption are solely given by
its role in indirectly taxing profits, i.e. intratemporal gains from taxing consumption, and amount to
19.88%. While both the intertemporal and the intratemporal gains from taxing consumption are sizeable,
the intratemporal gains are quantitatively more important.
In the second part of Table (4) we increase the initial debt equal to 120% of GDP, a value that is almost
twice as big as in the benchmark calibration. The results follow consistently as expected. Compared to
the benchmark, under both tax scenarios, the Ramsey Planner suffers a greater welfare loss and has a
stronger incentive to manipulate the initial bond price pbt . Moreover, under labour income taxation, the
higher need for fiscal revenues reinforce the substitution effect of labour supply. This pushes the Ramsey
planner to lower the government spending to income ratio compared to the benchmark case. In this case,
the gains from taxing consumption amount to 25.01%.
Table (5) presents the allocation and welfare under the alternative parametrisation where σ 6= 1. In
order to be consistent with the benchmark case, we re-calibrate κ and ωg so that in the efficient allocation
`sp = 0.7 and gsp
hsp = 0.2. We keep all the remaining parameters at their benchmark values. The literature
has not reached a unanimous consensus on the EIS. To this end, we solve the model with σ = 2, which is
consistent with low values of EIS as typically found by the micro literature, see among others Blundell,
Browning, and Meghir (1994). We also solve the model with σ = 0.5, a value consistent with some macro-
finance literature that finds high values of the EIS, see among others Bansal and Yaron (2004). Given
this parametrisation, the utility function is not longer separable in consumption and leisure. This means
among other things, that labour supply decisions directly affect the marginal utility of consumption, and
in turn, consumption decisions directly affect the marginal utility of leisure.
Despite this, the main message from this experiment does not change qualitatively from the benchmark
17
Table 4: Tax Rates and Allocations, Public Debt
First Best γb = 1.2 τh = 0Variable t = 0 t ≥ 1 t = 0 t ≥ 1
γb = 0, κ = 6.944, ωg = 0.25Consumption tax rate - 0 0 0.250 0.250Labour income tax rate - 0.229 0.229 0 0Hours worked, Income 0.300 0.239 0.239 0.245 0.249Leisure 0.700 0.761 0.761 0.755 0.755Consumption 0.240 0.193 0.193 0.196 0.196Government Spending 0.060 0.046 0.046 0.049 0.049Public debt - 0 0 0 0Public debt price - 0.979 0.979 0.979 0.979Consumption-income ratio 0.800 0.809 0.809 0.800 0.800Public spending-income ratio 0.200 0.191 0.191 0.200 0.200Public debt-income ratio - 0 0 0 0Welfare-eq. consumption loss 0 0.058 0.046
γb = 1.2, κ = 6.944, ωg = 0.25Consumption tax rate - 0 0 0.535 0.272Labour income tax rate - -0.065 0.262 0 0Hours worked, Income 0.300 0.273 0.234 0.222 0.243Leisure 0.700 0.727 0.766 0.779 0.757Consumption 0.240 0.234 0.189 0.177 0.194Government Spending 0.060 0.039 0.044 0.044 0.049Public debt - 0.327 0.315 0.265 0.200Public debt price - 1.210 0.979 1.075 0.979Consumption-income ratio 0.800 0.858 0.820 0.800 0.800Public spending-income ratio 0.200 0.142 0.190 0.200 0.200Public debt-income ratio - 1.200 1.345 1.200 0.823Welfare-eq. consumption loss 0 0.068 0.051
case. However There are some important quantitative differences. With a low EIS, i.e. σ = 2, the
optimal long run level of public spending is higher than the benchmark case and government spending-to-
income ratio is higher under consumption taxation than under labour income taxes. Sustaining a bigger
government size requires higher taxes in the long run. In turn, higher taxes depress labour supply and
consumption, thus decreasing the overall welfare. Interestingly, under labour income taxation, the need
for higher taxes in long-run generates a stronger incentive to manipulate the initial bond prices, i.e. pb0 is
bigger when σ = 2 than in the benchmark case. This is due to the fact that with this tax component and
a low EIS, it is more difficult to affect initial private consumption through the Euler equation. Differently,
under consumption taxation the Ramsey planner has a weaker incentive to manipulate the initial bond
prices compared with the benchmark case. This is because contrary to the labour income tax case, the
Ramsey planner can now influence directly the Euler equation via a dynamic pattern of consumption
18
taxes in period zero and in subsequent periods.
Consistently, all these results are reversed when the model is solved with a high EIS. Compared to
the benchmark case, the government size is now smaller, taxes are lower and the welfare loss is greatly
reduced under both tax scenarios. Remarkably the welfare gains from taxing consumption are very similar
across the different EIS experiments and amount to 22.44% when σ = 2 and to 22.82% with σ = 0.5.
Table 5: Tax Rates and Allocations, Non-Separable Preferences
First Best τ c = 0 τh = 0Variable t = 0 t ≥ 1 t = 0 t ≥ 1
σ = 2, κ = 5.050, ωg = 2.204Consumption tax rate - 0 0 0.292 0.2767Labour income tax rate - -0.026 0.247 0 0Hours worked, Income 0.300 0.256 0.218 0.221 0.226Leisure 0.700 0.744 0.782 0.778 0.773Consumption 0.240 0.220 0.177 0.180 0.181Government Spending 0.060 0.035 0.042 0.042 0.046Public debt - 0.164 0.154 0.142 0.130Public debt price - 1.325 0.979 1.004 0.979Consumption-income ratio 0.800 0.862 0.809 0.812 0.799Public spending-income ratio 0.200 0.138 0.191 0.188 0.201Public debt-income ratio - 0.640 0.707 0.640 0.575Welfare-eq. consumption loss 0 0.087 0.067
σ = 0.5, κ = 8.547, ωg = 0.096Consumption tax rate - 0 0 0.461 0.261Labour income tax rate - 0.152 0.248 0 0Hours worked, Income 0.300 0.265 0.256 0.249 0.261Leisure 0.700 0.734 0.744 0.751 0.738Consumption 0.240 0.218 0.207 0.197 0.209Government Spending 0.060 0.048 0.049 0.052 0.052Public debt - 0.170 0.180 0.159 0.111Public debt price - 1.021 0.979 1.018 0.979Consumption-income ratio 0.800 0.820 0.808 0.792 0.800Public spending-income ratio 0.200 0.180 0.192 0.208 0.200Public debt-income ratio - 0.640 0.705 0.640 0.425Welfare-eq. consumption loss 0 0.042 0.032
Table (6) presents the results from varying the Frisch elasticity of labour supply. This exercise is
particularly important as the micro and macro literature tend to differ on the estimates of this structural
parameter. For this reason we solve the model with two values that are somehow at the extremes of the
available estimates of the elasticity of labour supply. In particular, we consider (i) ϕ = 0.5, which is in line
with recent micro estimates such as Domeij and Floden (2006) (see also Guner, Kaygusuz, and Ventura,
2012), and (ii) ϕ = 5 as a high value, which is sometimes chosen to better match the intertemporal
19
variation of aggregate hours (e.g. Galı, Lopez-Salido, and Valles, 2007). We adjust κ appropriately in
each case as described for the benchmark parametrisation. When the Frisch elasticity is low, i.e. ϕ = 0.5,
the substitution effect of labour supply is reduced. Thus the policy-maker can sustain the same level
of public expenditure and debt at a lower welfare cost. Under labour income tax this implies that the
Ramsey Planner can afford to push the public spending-income ratio closer to the first best level without
distorting excessively the economy. Of course this result is reversed when the Frisch elasticity of high i.e.
ϕ = 5 as the substitution effect of labour supply gets bigger. Differently, under consumption taxation, the
policy-maker finds it optimal to keep the government spending to income ratio at the first best regardless
of the labour supply elasticity, see Proposition 5. Therefore the economy is facing a higher welfare loss
when the Frisch elasticity is high compared to when the Frisch elasticity is low. The gains from taxing
consumption are roughly the same under the two different parametrisations and amount to around 22.4%.
Table 6: Tax Rates and Allocations, Frisch Elasticity
First Best τ c = 0 τh = 0Variable t = 0 t ≥ 1 t = 0 t ≥ 1
ϕ = 0.5, κ = 15.432, ωg = 0.25Consumption tax rate - 0 0 0.444 0.261Labour income tax rate - 0.133 0.250 0 0Hours worked, Income 0.300 0.266 0.256 0.250 0.261Leisure 0.700 0.734 0.744 0.750 0.739Consumption 0.240 0.220 0.206 0.200 0.209Government Spending 0.060 0.046 0.049 0.050 0.052Public debt - 0.170 0.179 0.160 0.113Public debt price - 1.045 0.979 1.071 0.979Consumption-income ratio 0.800 0.827 0.806 0.800 0.800Public spending-income ratio 0.200 0.172 0.193 0.200 0.200Public debt-income ratio - 0.640 0.699 0.640 0.433Welfare-eq. consumption loss 0 0.041 0.032
ϕ = 5, κ = 4.418, ωg = 0.25Consumption tax rate - 0 0 0.311 0.265Labour income tax rate - -0.039 0.241 0 0Hours worked, Income 0.300 0.252 0.201 0.206 0.212Leisure 0.700 0.748 0.798 0.794 0.789Consumption 0.240 0.215 0.164 0.164 0.169Government Spending 0.060 0.036 0.037 0.041 0.042Public debt - 0.161 0.160 0.132 0.123Public debt price - 1.283 0.979 0.985 0.979Consumption-income ratio 0.800 0.856 0.815 0.800 0.800Public spending-income ratio 0.200 0.144 0.185 0.200 0.200Public debt-income ratio - 0.640 0.795 0.640 0.582Welfare-eq. consumption loss 0 0.107 0.083
20
Finally, in Table (7) we explore the possibility that public and private consumption are complements
or substitutes in the Edgeworth sense. This is achieved by varying α ∈ [−0.5, 0.5], whereas a positive
value implies substitutability and a negative one complementarity. In the former case, an increase in
government spending decreases the marginal utility of consumption. This weakens the income effect of
labour supply and the increases the EIS. Ceteris paribus, a lower income effect means that the substitution
effect becomes relatively stronger. Compared to the benchmark case, this implies a stronger distortionary
effect of taxation that in turn pushes the Ramsey Planner to reduce government spending-to-income ratio.
Under labour income tax, the incentive to reduce public consumption is more pronounced than under
consumption taxation, although under both fiscal scenarios the government spending-to-income ratio fall
short of its first best. Furthermore, given the substitutability between private and public consumption,
the reduction of public spending is compensated by the increase in private consumption and this leads to
a lower welfare loss. The gains from taxing consumption reduces to 17.81%.
Conversely, with α = −0.5, an increase in government spending increases the marginal utility of
consumption. This weakens the income effect of labour supply and the increases the EIS. A higher
income effect decreases the distortionary effects of taxation thus pushing the Ramsey Planner to increase
government spending-to-income ratio. This ratio is only marginally higher under consumption taxation,
and under both fiscal scenarios the government spending-to-income ratio is higher than their first best
counterparts. In turn the lower allocation of resources to private consumption implies, compared to the
benchmark a higher welfare loss and a higher welfare gain from taxing consumption (24.22%).
4 Concluding Remarks
Consumption taxation has been widely used by governments around the globe as a source of fiscal revenues.
For example, as of August 2016, the value-added tax on standard items ranges from 17 (Luxembourg)
to 27 (Hungary) percent in European Union countries. We consider a simple economy characterised by
21
Table 7: Tax Rates and Allocations, Complementarity
First Best τ c = 0 τh = 0Variable t = 0 t ≥ 1 t = 0 t ≥ 1
α = 0.5, κ = 6.173, ωg = 0.111Consumption tax rate - 0 0 0.350 0.199Labour income tax rate - 0.059 0.202 0 0Hours worked, Income 0.300 0.261 0.241 0.233 0.247Leisure 0.700 0.739 0.758 0.767 0.753Consumption 0.240 0.226 0.205 0.196 0.208Government Spending 0.060 0.035 0.037 0.037 0.039Public debt - 0.167 0.177 0.149 0.113Public debt price - 1.068 0.979 1.039 0.979Consumption-income ratio 0.800 0.864 0.847 0.842 0.842Public spending-income ratio 0.200 0.136 0.153 0.158 0.158Public debt-income ratio - 0.640 0.735 0.640 0.456Welfare-eq. consumption loss 0 0.055 0.045
α = −0.5, κ = 7.936, ωg = 0.429Consumption tax rate - 0 0 0.391 0.281Labour income tax rate - 0.070 0.271 0 0Hours worked, Income 0.300 0.257 0.237 0.235 0.245Leisure 0.700 0.742 0.763 0.765 0.755Consumption 0.240 0.212 0.188 0.185 0.193Government Spending 0.060 0.045 0.050 0.050 0.052Public debt - 0.165 0.170 0.150 0.125Public debt price - 1.148 0.979 1.020 0.979Consumption-income ratio 0.800 0.824 0.789 0.789 0.789Public spending-income ratio 0.200 0.176 0.210 0.211 0.211Public debt-income ratio - 0.640 0.719 0.640 0.510Welfare-eq. consumption loss 0 0.062 0.047
infinitely lived representative households with love for public goods, a perfect competitive labour market
and a monopolistic competitive goods market. In this general environment, our analysis shows that
taxing consumption, compared to taxing labour income, allows the policy-maker to sustain both higher
public spending and higher aggregate welfare. We are able to disentangle two effects through which
consumption taxation overtake income taxes. First, consumption taxation serves as an indirect taxation
of the inefficient profits produced by the non competitive markets. This beneficial effect increases with the
assumed firms’ market power and disappears under perfect competition. Second, given the intertemporal
structure of consumption behaviour, taxing consumption allows the policy-maker to directly manipulate
households’ Euler equation in order to favourably influence the price of public debt. This effect increases
with the size of public debt. We we find that both effects are quantitatively important in an economy
calibrated to replicate some salient features of the US economy.
22
References
Adam, K. (2011, January). Government debt and optimal monetary and fiscal policy. European EconomicReview 55 (1), 57–74.
Bansal, R. and A. Yaron (2004). Risks for the Long Run: A Potential Resolution of Asset Pricing Puzzles.The Journal of Finance 59 (4), 1481–1509.
Blundell, R., M. Browning, and C. Meghir (1994). Consumer Demand and the Life-Cycle Allocation ofHousehold Expenditures. Review of Economic Studies 61 (1), 57–80.
Coleman, W. J. I. (2000). Welfare and Optimum Dynamic Taxation of Consumption and Income. Journalof Public Economics 76 (1), 1–39.
Correia, I. (2010, September). Consumption Taxes and Redistribution. American Economic Re-view 100 (4), 1673–94.
Correia, I., E. Farhi, J. P. Nicolini, and P. Teles (2013). Unconventional Fiscal Policy at the Zero Bound.American Economic Review 103 (4), 1172–1211.
Domeij, D. and M. Floden (2006). The Labor-Supply Elasticity and Borrowing Constraints: Why Esti-mates are Biased. Review of Economic Dynamics 9 (2), 242–262.
Farhi, E., G. Gopinath, and O. Itskhoki (2014). Fiscal Devaluations. Review of Economic Studies 81 (2),725–760.
Galı, J., J. D. Lopez-Salido, and J. Valles (2007). Understanding the Effects of Government Spending onConsumption. Journal of the European Economic Association 5 (1), 227–270.
Guner, N., R. Kaygusuz, and G. Ventura (2012). Taxation and Household Labour Supply. Review ofEconomic Studies 79 (3), 1113–1149.
Kimball, M. S. and M. D. Shapiro (2008, July). Labor supply: Are the income and substitution effectsboth large or both small? Working Paper 14208, National Bureau of Economic Research.
Klein, P., P. Krusell, and J.-V. Rıos-Rull (2008). Time-Consistent Public Policy. Review of EconomicStudies 75 (3), 789–808.
Lucas, R. J. and N. L. Stokey (1983). Optimal fiscal and monetary policy in an economy without capital.Journal of Monetary Economics 12 (1), 55–93.
Martin, F. M. (2010). Markov-perfect capital and labor taxes. Journal of Economic Dynamics andControl 34 (3), 503 – 521.
Motta, G. and R. Rossi (2013). Ramsey monetary and fiscal policy: the role of consumption taxation.Technical report.
Natvik, G. J. (2009, 02). Government Spending and the Taylor Principle. Journal of Money, Credit andBanking 41 (1), 57–77.
Ni, S. (1995). An empirical analysis on the substitutability between private consumption and governmentpurchases. Journal of Monetary Economics 36 (3), 593 – 605.
23
Rotemberg, J. J. and M. Woodford (1998, May). An optimization-based econometric framework for theevaluation of monetary policy: Expanded version. Working Paper 233, National Bureau of EconomicResearch.
Trabandt, M. and H. Uhlig (2011). The Laffer curve revisited. Journal of Monetary Economics 58 (4),305–327.
A Proof of Proposition 4
For the case of log utility, the FOCs for t ≥ 1 of the Ramsey specialise to
u`,t (1 + γt)− γt[u``,t (1− `t)−
1
εuc,t
]= ug,t, (21)
uc,t (1 + γt) + γtucc,tct − γtucc,t1
ε(1− `t) = ug,t. (22)
Then using the fact in the log case −ucc,tct = uc,t, equation (22) can be written as
uc,t − γtucc,t1
ε(1− `t) = ug,t. (23)
Given that γt > 0 and ucc,t < 0, equation (23) implies that ug,t ≥ uc,t. Then we can write (4) asu`,t = wt
(1− τht
)uc,t. Using that wt
(1− τht
)≤ 1, it follows that uc,t ≥ u`,t, which implies ug,t ≥ u`,t.
Clearly, if ε→∞, equation (23) implies ug,t = uc,t. In this case, given the log preferences in governmentspending, gt/yt = gspt /y
spt . QED
B Proof of Proposition 5
Similarly, under consumption taxation, the Ramsey planner for t ≥ 1 collapses to
u`,t (1 + γt)− γt (1− `t)u``,t = ug,t, (24)
uc,t + γtε− 1
ε[ucc,tct + uc,t] = ug,t. (25)
Then using again that −ucc,tct = uc,t, it follows from (25) that uc,t = ug,t. Then, the log preferences ingovernment spending imply that gt/yt = gspt /y
spt . QED
24